McKinsey’s 2026 State of AI survey finds that artificial intelligence has become a routine tool at many companies: nearly nine in ten respondents said their organization regularly uses AI in at least one business function, and 44% reported scaling AI at the enterprise level — up from 38% in 2025. The share of organizations using AI in at least three functions also rose, from 51% to 56%.
One notable finding is that AI is reshaping the traditional buy-or-build decision. Thirty-two percent of respondents said their organization had already decided not to purchase at least one piece of software or software feature because the capability could be built in-house using AI coding tools. This effect is strongest at technology companies (41%), followed by healthcare firms (39%), and is also pronounced in professional services and the energy and materials sector (both 38%).
McKinsey notes that AI-assisted programming can do more than raise developer productivity: it can reallocate enterprise IT budgets by prompting companies to weigh more carefully which technologies to buy, which to build internally, and where to invest in in-house integration capabilities.
Employee productivity gains don’t yet equal corporate profit boosts
The survey highlights a sharp contrast: AI appears to deliver clear benefits at the individual level but far more limited effects on corporate financials. Eighty percent of employees who use AI at work said it improved their productivity, and about half reported making better decisions with AI. Yet only 37% said AI had positively contributed to their company’s operating profit (EBIT), and that proportion has remained largely unchanged since 2025.
McKinsey does identify areas with measurable economic impact: respondents most often cited cost reductions in supply chain, service operations, and manufacturing. Revenue gains were most frequently linked to AI use in marketing and sales, product and service development, and software development.
Running AI carries costs too
Wider adoption has also revealed that operating AI can be expensive: 20% of respondents said their organization has limited AI use because of operating costs, including token costs associated with model usage. Despite cost pressure, firms are not generally pulling back on investment: 28% said their organization already spends more than 10% of its total IT and communications budget on AI technologies, and 60% expect to increase AI investment over the next year.
Michael Chui, McKinsey’s lead analyst, argues that “tokenomics” is becoming an increasingly important concept. Even if per-token costs fall, the volume of tokens consumed and generated may grow faster as more complex reasoning tasks and AI assistants proliferate, forcing companies to evaluate not just what AI can do but whether it is cost-effective for a given task.
Only a small elite is realizing substantial AI-driven profits
Only 6% of respondents belonged to what McKinsey calls “AI high performers” — organizations where AI generated at least a 5% EBIT impact and where respondents judged the technology’s value creation to be significant. That share did not increase from 2025.
The gap is not simply explained by larger AI budgets. High-performing firms are 3.3 times more likely to plan to fundamentally transform their business with AI over the next three years. Nearly three-quarters have already redesigned workflows because of AI, compared with about one in four among other organizations. High performers are twice as likely to scale software-coding assistants and lean more toward building rather than buying: nearly half have decided not to purchase certain software or features because they can produce them in-house with AI, versus 31% among other firms.
Tara Balakrishnan, a McKinsey associate partner, says organizational adaptability — not the technology itself — is becoming the main bottleneck to AI adoption.
Implications for the software and consulting markets
McKinsey’s data suggests that if AI coding systems make it cheaper and faster to create bespoke enterprise functions, off-the-shelf software may become less attractive in some cases. The survey does not show mass exit from the enterprise software market, but it does indicate that AI is starting to rewrite a fundamental IT decision: what to buy versus what to build.
AI is also disrupting the consulting industry’s traditional business model by automating routine tasks such as data collection, analysis, presentation preparation, and document processing. In the UK, “AI-native” consultancies have emerged that try to compete with large firms such as Deloitte, PwC, EY, and KPMG using smaller teams augmented by AI agents. The technology lowers barriers to entry and calls into question the pyramid model that relies on many juniors and billable hours. Pricing may shift away from hourly billing toward outcome-based and subscription models for tasks an AI system can perform in a fraction of prior time.
Nevertheless, large consultancies retain advantages from global networks, client relationships, and multibillion-dollar AI investments; mid-sized firms may be most vulnerable if they lack both deep pockets and the agility of startups.
Overall, McKinsey’s research indicates that while AI is already reshaping enterprise IT decisions and supplier markets, broad-based profit improvement has so far been concentrated in a narrow group of firms. Future success will depend heavily on companies’ ability to reorganize workflows, structures, and operating models around the technology.
